Welcome to ChromBERT-tools¶
ChromBERT-tools is a lightweight toolkit built on top of the ChromBERT foundation model. It helps users generate, model, and interpret context-specific regulatory representations from their own genomic data through command-line interfaces (CLIs) and Python APIs.
ChromBERT is a pre-trained foundation model designed to capture genome-wide co-association patterns among transcription regulators and to learn context-specific transcriptional regulatory networks.
Key features¶
Simple interface — provides a single
chrombert-toolsentry point with composable subcommands.Human and mouse support — supports hg38 and mm10 genomes at multiple resolutions.
Modular workflows — supports representation generation, predictive modeling, and regulatory interpretation.
Cell-type-specific analysis — supports context-specific analyses using accessibility tracks or fine-tuned checkpoints.
CLI and Python API — exposes core functions through both shell commands and importable Python APIs.
Core functionality¶
ChromBERT-tools is organized into three functional layers:
Representation generation — generate context-specific representations for genomic regions or regulators.
Predictive modeling — train or apply models for region classification, region activity prediction, and gene activity prediction.
Regulatory interpretation — infer functional relationships among regions, regulators, and regulatory programs.
It also provides end-to-end application workflows for common biological tasks, including cell-type-specific key regulator identification, driver regulator identification during cell-state transitions, context-specific cofactor analysis, and TF-binding prediction.
Getting started¶
Follow Installation to install ChromBERT-tools using the recommended Apptainer image or from source.
Download the required ChromBERT model and annotation files using
download-data.See Usage for the full list of available commands and tutorials.
To list all available commands:
chrombert-tools --help
To get help for a specific command:
chrombert-tools <command> --help
User guide¶
Links¶
ChromBERT GitHub: https://github.com/TongjiZhanglab/ChromBERT
ChromBERT-tools GitHub: https://github.com/TongjiZhanglab/ChromBERT-tools
Documentation: https://chrombert-tools.readthedocs.io/
Contact¶
For questions or suggestions, please contact us at 2211083@tongji.edu.cn.